Computing models for quotients of modular curves (Q2045741)

This article describes an algorithm which computes models for modular curves associated to general congruence subgroups, as well as quotients of these curves by certain automorphism groups. More precisely, let \(G\) be a subgroup of \(\mathrm{GL}_2(\mathbb{Z}/N\mathbb{Z})\), and let \(X_G\) be the associated modular curve over \(\mathbb{Q}\). The algorithm computes a model of \(X_G\) in the case \(\mathrm{det}(G) = (\mathbb{Z}/N\mathbb{Z})^*\) (that is, the modular curve is geometrically connected), \(-I \in G\), and \(G\) is normalised by the matrix \(\begin{pmatrix} 1 & 0 \\ 0 & -1 \end{pmatrix}\). The algorithm also computes models for the quotient of \(X_G\) by finite groups of automorphisms coming from the normaliser of the congruence subgroup in \(\mathrm{PGL}_2^+(\mathbb{Q})\), under mild assumptions. This includes the case of the Atkin-Lehner involutions. To achieve this, the author presents a new algorithm which computes the space of cusp forms of weight 2 on the given group. So far, algorithms for computing cusp forms have been implemented in Magma, Sage and PARI/GP only for the groups \(\Gamma_0(N)\) and \(\Gamma_1(N)\) (and the intermediate subgroups). To deal with the principal congruence subgroup \(\Gamma(N)\), the trick is that the transformation \(z \mapsto Nz\) conjugates \(\Gamma(N)\) to the group \(\Gamma_0(N^2) \cap \Gamma_1(N)\). The algorithm in this article uses modular symbols, but also, importantly, the so-called twist operators, an idea due originally to Cremona. Once a basis of the space of cusp forms has been computed (in the form of \(q\)-expansions), equations for the model can then be obtained using the canonical embedding when the curve is not hyperelliptic, as in Galbraith [Equations for modular curves. Oxford (PhD thesis) (1996)]. The algorithm has not been implemented yet. However, it has been used by the author to compute models of several modular curves of level 35. These curves are involved in the proof of modularity results for elliptic curves over totally real number fields [\textit{N. Freitas} et al., Invent. Math. 201, No. 1, 159--206 (2015; Zbl 1397.11086); \textit{M. Derickx} et al., Algeb. Number Theory, 14, 7, 1791--1800 (2020; Zbl 1471.11178); \textit{J. Box}, ``Elliptic curves over quartic fields not containing \(\sqrt{5}\) are modular'', Preprint, \url{arXiv:2103.13975}]. The article also contains a detailed study of morphisms between modular curves and their moduli interpretation, enabling in particular to decide their fields of definition.